Perceptions of Political Corruption in Latin American Democracies

Damarys Canache, Michael E. Allison

Latin American Politics & Society, Volume 47, Number 3, Fall 2005,
pp. 91-111 (Article)
Published by University of Miami
DOI: 10.1353/lap.2005.0031

For additional information about this article

http://muse.jhu.edu/journals/lap/summary/v047/47.3canache.html

Access provided by University Diego Portales (24 Sep 2013 12:02 GMT)

Perceptions of Political Corruption

in Latin American Democracies
Damarys Canache
Michael E.Allison
ABSTRACT
Political corruption poses a serious threat to the stability of developing democracies by eroding the links between citizens and governments. Using data on national levels of corruption (Transparency
International 1997 CPI index) and individual opinion (199597
World Values Survey), this study finds that Latin Americans are quite
aware of the seriousness of corruption in their countries. The ensuing question is whether citizens can connect their views about corruption to appraisals of their authorities and institutions and of
democracy more generally. Collectively, the findings suggest that
they can, and that the necessary ingredients for accountability are
present in Latin America. The possible dark side of mass opinion
on corruption is that pervasive misconduct may poison public sentiment toward democratic politics. On this score, the analysis found
that this attitude affected only support for specific administrations
and institutions.

olitical corruption is a severe problem facing many democratic

nations. At its most basic, political corruption involves the abuse of
public power for some private benefit. Its existence may distort government outputs because actors involved in corrupt practices gain disproportionate benefits from government. This, in turn, may distort democratic procedures, because policies result not from an open clash of
ideas in the marketplace but instead from back-alley deals.
Political corruption is an especially significant problem in Latin America, where corruption scandals tarnished many Latin American governments in the 1990s. In Brazil, President Fernando Collor de Mello was
forced to resign following a chain of scandals involving an influence-peddling ring during the electoral campaign, redirection of public funds by
his wife, and reports of a lavish private life. In Venezuela, Carlos Andrs
Prez was suspended from the presidency in the midst of impeachment
proceedings that were initiated following news of his misappropriation of
public funds. In Ecuador, the media uncovered evidence that President
Abdal Bucaram stashed away for himself money that had been raised
during a Christmas telethon to help the poor. This triggered a political
process ending with his removal by Congress for mental incapacity
(Stapenhurst 2000). In Mexico, the reputation of former president Carlos
91

92

LATIN AMERICAN POLITICS AND SOCIETY

47: 3

Salinas de Gortari was severely tarnished largely by the actions of his

now-imprisoned brother Ral, who was accused of murdering a political
opponent, playing a major role in drug trafficking, and stealing more than
$100 million. Carlos Salinas has lived in exile in Ireland since he left the
Mexican presidency. In Nicaragua, former president Arnoldo Alemn has
faced fraud, embezzlement, and money-laundering charges. In Argentina,
President Carlos Menem left office in 1999 following a spate of scandals.
Two years later, he was arrested on charges of illegal arms trafficking and
accepting millions of dollars in bribes.
Much of the research conducted on corruption and its political
impact thus far has examined matters such as the factors that induce corruption to take root in a nations political system; its systemic negative
consequences, such as clientelism and cronyism (see, for example, Banfield 1958; Johnston 1979; Etzioni-Halevy 1985); its systemic positive
consequences, such as political stability (Huntington 1969; Waterbury
1976; Becquart-Leclerq 1989; Heidenheimer et al. 1989); and its consequences for policymaking, especially for economic policy (Tulchin and
Espach 2000). The focus of this study, in contrast, is the link between
corruption and the perceptions and attitudes of a nations citizens. In
exploring the significance of corruption for mass politics, this study follows a path that so far has been traveled by only a relative handful of
scholars (for example, Morris 1991; Shin 1999; Camp et al. 2000; Della
Porta 2000; Pharr 2000; Seligson 2002).1
Political corruption raises several issues regarding the relationship
between citizens and their governments. At one level, when the highest
authorities are frequently implicated in electoral manipulation, financial
scandals, or the abuse of public resources to achieve personal benefits,
their bases of authority and legitimacy may be seriously undermined. At
another level, corruption constitutes a violation of the unwritten contract
between citizen and public official. Political corruption distorts public
demands, increases the cost, reduces quality, and delays the completion
of public works, delays access to public administration for those people
who do not pay bribes and reduces the productivity of all those
involved with the state (Della Porta 2000). Thus it is no surprise that
some scholars view corruption as a common and profound obstacle to
the consolidation of new democracies (Schedler et al. 1999, 1).
A vital step in combating political corruption is that citizens be able
to hold leaders accountable when they engage in malfeasance. For this
vertical accountability (ODonnell 1999) to exist, citizens need to be
able to hold political leaders to some standard and punish them when
they violate that standard. Accountability of this sort entails two additional requirements. First, citizens must accurately perceive corruption
when it does exist. If citizens simply believe that all politicians are corrupt, this would most likely bode poorly for the fate of democracy, as

CANACHE AND ALLISON: CORRUPTION

93

none of the electoral alternatives would appeal to citizens. Additionally,

there would be nothing left to explain if citizens from all countries
viewed their democratic leaders as either all being corrupt or not corrupt at all. We find that possibility unlikely.
Second, citizens need to adjust their opinions of the relevant leaders (or institutions) accordingly. Thus, when citizens accurately perceive
corruption and they adjust their opinions of the leader, they will be
better able to hold that leader accountable through democratic
processes (elections, judicial procedures, protest, and so on). Positive
consequences can result when citizens identify certain leaders or political institutions as corrupt and they are able to hold them accountable
through various forms of sanction. When political leaders are thought to
be corrupt and the legislature or citizens hold the leader accountable,
this is an example of democracy in action. However, if citizens misplace
blame, it is possible that corruption can have a destabilizing affect on
democratic political systems. For instance, if citizens believe that corruption is an inherent component of a democratic regime and not
merely attributable to an individual political leader or institution, this
may weaken support for democracy as a form of governance, thereby
increasing the risk that an alternate regime will emerge. Therefore, it is
critically important to identify whether citizens can accurately assess the
level of corruption in their nation and whether perceptions of corruption affect their support for the incumbent government, the political
system and democracy as a system of governance.
This article tries to disentangle the relationship between political
corruption and public opinion in several Latin American nations using
data from the 199597 World Values Survey. The World Values Survey is
a worldwide survey project that investigates social, cultural, and political attitudes. The third wave of this project, conducted from 1995
through 1997, was performed in national representative samples of 54
independent nations. Specifically, it assesses the characteristics of citizens perceptions of corruption and considers whether those perceptions affect support for incumbent political officials, political institutions,
and democracy itself, broadly speaking. The analyses begin with a brief
review of the nature of corruption in Latin America.

THE SIGNIFICANCE
IN LATIN AMERICA

OF

CORRUPTION

Over the last several years, political events in both North and South have
brought a renewed vigor to the study of political corruption. In the
advanced industrial democracies, corruption scandals have brought down
several prominent political figures in Italy (the clean hands investigation), Germany (Helmut Kohl), the United States (Rep. James Traficant, D-

94

LATIN AMERICAN POLITICS AND SOCIETY

47: 3

Ohio), and Great Britain (Peter Mandelson). Although these events have
undoubtedly undermined citizen trust and confidence in the government,
they have not destroyed the democratic system in those countries. Accusations and incidents of political corruption in the global South, by contrast, have shaken many new democratic regimes to their core.
Recent history in Latin America illustrates both the extent of corruption at the national level and the difficulties of combating it. Since
the early 1990s, nine Latin American presidents or former presidents
have faced judicial proceedings or have been dismissed on corruption
charges.2 One indicator of the severity of corruption in Latin America is
provided by Transparency Internationals Corruption Perception Index
(CPI). Table 1 depicts CPI scores from 1997 and 2003 for all available
Latin American nations and advanced democracies. The 1997 data contributed to this studys multivariate analyses involving the 199597
World Values Survey. The 2003 data are the most recent indicators available; and comparison of 1997 and 2003 scores reveals that CPI scores
for a given nation typically vary little in the short run.
As coded here, scores on the CPI range from zero (least corrupt) to
10 (most corrupt). In the advanced democracies, the mean CPI score in
both years is under 2.0.3 Although moderately high corruption is indicated in nations such as Belgium and Italy, the news in the advanced
democracies is quite good overall. A starkly different story emerges in
Latin America. Among Latin American nations, the mean CPI value
exceeds 6.0 in both 1997 and 2003, and only Chile receives CPI scores
that would be inconspicuous in advanced democracies.
A few words are in order concerning the measure of corruption
used in table 1 and throughout the following analyses. The CPI is a joint
initiative of Gttingen University and Transparency International, and
has become a popular measure in cross-national statistical analyses over
the last several years. The CPI is an attempt to assess the level of corruption in a country by compiling a poll of polls based on the perceptions of both national and foreign business leaders and risk analysts
working for multinational firms and institutions in each country. Transparency International defines corruption as the misuse of public power
for private benefits; for example, the bribing of public officials, taking
kickbacks in public procurement, or embezzling public funds (Graf
Lambsdorff 2003). It uses this definition of corruption to seek out polls
asking comparable questions in countries around the world. Transparency International calculates CPI scores for a nation if data from a
minimum of four surveys are available.
Although there is no purely objective measure of corruption, the
CPI scores provided by Transparency International offer a relatively
accurate depiction of the level of corruption in each country. The
responses of business people and risk analysts, whose job it is to take

CANACHE AND ALLISON: CORRUPTION

95

Table 1. Corruption Levels in Selected Democracies

Advanced Democracies
_____________________________
1997
2003

Latin American Democracies

_____________________________
1997
2003

Austria
Australia
Belgium
Canada
Denmark
Finland
France
Germany
Italy
Japan
Luxembourg
Netherlands
New Zealand
Norway
Sweden
Switzerland
United Kingdom
United States

2.4
1.1
4.8
0.9
0.1
0.5
3.3
1.8
5.0
3.4
1.4
1.0
0.8
1.1
0.6
1.4
1.8
2.2

2.0
1.2
2.4
1.3
0.5
0.5
3.1
2.3
4.7
3.0
1.3
1.1
0.5
1.2
0.3
1.2
1.3
2.5

Argentina
Bolivia
Brazil
Chile
Colombia
Costa Rica
Ecuador
El Salvador
Guatemala
Honduras
Mexico
Nicaragua
Panama
Paraguay
Peru
Uruguay
Venezuela

7.2
8.0
6.4
3.9
7.8
3.5

7.3

5.6
7.2

7.5
7.7
6.1
2.6
6.3
5.7
7.8
6.3
7.6
7.7
6.4
7.4
6.6
8.5
6.3
4.5
7.6

Average

1.9

1.8

Average

6.3

6.6

Source: Transparency International 1997, 2003.

into consideration the level of corruption in a country before deciding

whether and how much to invest in the countrys markets, yield a measure that is mostly the result of informed elitebased surveys. The CPI
index is widely used by economists assessing the effect of corruption on
macroeconomic variables, and it is no doubt the best overall indicator
of national levels of corruption worldwide (Seligson 2002, 415).
Indeed, we believe that this is a more accurate measure of corruption
than other potential sources. Corruption measures based on news stories or judicial proceedings are in many ways more a reflection of investigative journalism, freedom of the press, and an effective and independent judiciary than an indication of the level of corruption. Although
CPI scores do retain an element of subjectivity, they are the best available indicators for our purposes.4
The CPI data reported in table 1 reveal that political corruption is
widespread in most Latin American nations, particularly when those
nations are viewed in the context of advanced democracies. This claim
will be unsurprising for even casual observers of the region. But what

96

LATIN AMERICAN POLITICS AND SOCIETY

47: 3

connection exists between corruption in Latin America and citizens

political perceptions? Two central questions must be considered. First,
do mass perceptions of corruption share the critical tone reflected in the
CPI data? Second, to the extent that citizens do perceive some level of
corruption in Latin America, whom do they hold accountable? Most significantly, do they blame the various institutions and leaders of the state,
or do they blame democracy itself?

PERCEPTIONS

OF

CORRUPTION

If citizens are to hold leaders accountable for political corruption, the

initial step is to recognize the severity of the problem. Before citizens
decide to engage in popular protest or to punish corrupt leaders and
institutions at the ballot box, they first must characterize the degree of
corruption in their nation accurately.
CPI data indicate that the expert view holds that corruption is relatively high in most Latin American nations, although corruption is only
moderately severe in Costa Rica and Uruguay and comparatively minor
in Chile. In exploring the relationship between corruption and mass
opinion, this study initially looks at how much mass perceptions of corruption mirror the elite view indicated by the CPI. It answers this question with data from the 199597 World Values Survey (WVS) and the
1997 Transparency International CPI.5
The WVS includes data from seven Latin American nations: Argentina,
Brazil, Chile, Colombia, Mexico, Uruguay, and Venezuela. Collectively,
1997 CPI scores from these countries encompass nearly 40 percent of the
CPIs theoretical range. In 1997, corruption as reported on the CPI was
moderately low in Chile (3.9), moderate to moderately high in Uruguay
and Brazil, and high in Argentina, Venezuela, Mexico, and especially
Colombia (7.8). Thus, if citizens perceptions resemble those of the
experts whose views contribute to CPI scores, then opinion regarding the
severity of corruption should vary noticeably across these nations.
We can assess the nature of citizens perceptions at both the aggregate level and the individual level using the CPI scores as the basis of
comparison. At the aggregate level, we include CPI and WVS data from
both the seven Latin American nations and eight advanced democracies.
If aggregate mass perceptions are accurate, then the advanced democracies should have low mean levels of perceived corruption; medium
levels of perceived corruption should be found in Chile and Uruguay;
and high levels of perceived corruption should be found in the other
five Latin American nations.
The WVS measures perceived corruption with data from the item
How widespread do you think bribe taking and corruption is in this
country? Respondents were presented with four choice options, rang-

CANACHE AND ALLISON: CORRUPTION

97

Figure. 1. Correlation Between Country-level Corruption

(Transparency International-CPI) and Perceived Corruption

Source: Transparency International; World Values Survey 199597

ing from almost no public officials are engaged in it to almost all

public officials are engaged in it. We retain the original coding scheme,
with scale values ranging from 1 (low perceived corruption) to 4.
Figure 1 illustrates that aggregate perceived corruption does indeed
align closely with the expert view as represented by the CPI.6 The correlation between CPI values and mean perceived corruption scores is
0.91, suggesting a high level of aggregate perceptual accuracy. Among
the Latin American cases, those citizens living in the low to moderate
range of corruption according to the CPI (Chile and Uruguay) also collectively perceive the level of corruption to be in the same range. On
the other hand, those countries that scored more poorly on the CPI
measure (Argentina, Brazil, Colombia, Mexico, and Venezuela) were
perceived as quite corrupt by the average citizen. These results provide
evidence that there is a great deal of commonality between citizens perceptions of levels of corruption and the elite-level depiction of corruption provided by the CPI scale.
We also can assess the distribution of individual-level responses on
the corruption item. Of interest is what portion of respondents makes
what might be hesitantly labeled perceptual errors. For instance, we
might reasonably conclude that Chileans are off base if they report that
virtually all officials in their nation are engaged in corrupt practices, and

98

LATIN AMERICAN POLITICS AND SOCIETY

47: 3

Table 2. Perception of the Extent of Corruption

in Seven Latin American Countries (percent)
Public Officials Engaged in Corruption

Argentina (N = 1,032)
Brazil (N = 1,096)
Chile (N = 949)
Colombia (N = 2,910)
Mexico (N = 1,396)
Uruguay (N = 906)
Venezuela (N = 1,106)

Almost
all

Most

A few

Almost
none

37.27
58.94
17.24
44.40
44.13
14.37
44.03

41.18
23.54
27.47
32.96
30.87
30.07
30.74

18.50
16.79
50.91
19.79
20.63
52.34
22.88

3.07
0.73
4.38
2.85
4.37
3.23
2.85

Source: World Values Survey, 199597.

we might also look askance at Colombians who believe that few public
officials there have experienced the taint of corruption. Table 2 depicts
the relevant data. Although few respondents see almost no corruption in
their respective nations, responses are divided relatively evenly among
the other three response categories. In no country did the percentage of
respondents who claimed that almost no public officials are engaged in
it rise above 5 percent, and only in Chile and Uruguay, two of the
regions three most corruption-free states, did the percentage of respondents finding that few or no public officials are engaged in corruption
exceed 50 percent. On the other hand, over 70 percent of the citizens in
Argentina, Brazil, Colombia, Mexico, and Venezuela believe that most or
almost all public officials are corrupt. As for perceptual errors, only a relative handful of respondents are too generous in their assessments. On
the flip side, the question of whether respondents are too critical finds
that some people in Uruguay, and especially Chile, perceive the situation
in their respective nations as being much worse than the CPI indicates.
What explains the variance in the range of perceived corruption
seen in table 2? Three paths warrant consideration. First, individual-level
characteristics of WVS respondents may be associated with the predisposition to view corruption as high or low. Variables are included to
account for sex, age, education, social class, life satisfaction, interpersonal trust, and interest in politics. Second, variance in the actual conditions in each nation may affect respondents perceptions. Thus, using
CPI data as a surrogate to capture cross-national variance in corruption,
we are interested in whether Latin Americans individual-level judgments correspond with the CPI. Third, if a link between CPI scores and
mass perceptions of corruption is identified at the individual level, we
expect that the strength of this connection will be variable. Given that

CANACHE AND ALLISON: CORRUPTION

99

Table 3. Factors Affecting Level of Perceived Corruption

in Six Latin American Countries
(ordered logistic regression estimates)
Interactive
Main
Effect
Model
Effect
Model
___________________ __________________
Robust
Robust
Standard
Standard
Coefficient Errora Coefficient Errora
Sex
Age
Education
Social class
Satisfaction with life
Interpersonal trust
Political Interest
Transparency International
Score (nations
corruption level)
Transparency International
Score*Education
Transparency International
Score*Political Interest
Number of Observations
Model 2

0.137*
0.012**
0.033
0.119**
0.001
0.527**
0.187***

0.064
0.004
0.028
0.044
0.004
0.154
0.043

0.137*
0.012**
0.050
0.121**
0.001
0.527**
0.692***

0.063
0.004
0.127
0.043
0.003
0.171
0.192

0.423***

0.055

0.292**

0.101

0.003
0.079*
5,967
679.7

0.020
0.031

5,967
691.46

***p < .001 **p <.01 *p < .05 #p < .10

aAdjusted for clustering on country.
Dependent variable: level of perceived corruption, a four-category ordinal measure.
Notes: For tables 35, data from Argentina, Brazil, Chile, Mexico, Uruguay, and
Venezuela. See appendix for wording and descriptive statistics.
Source: World Values Survey, 199597; Transparency International.

the CPI is an elite-level indicator, we surmise that the correspondence

between CPI scores and mass perceptions will be greatest for those
respondents who are highly interested in politics and who have the
highest levels of education.
Because our dependent variable is a four-category ordinal indicator,
we assess the determinants of perceived corruption using ordered logistic regression. Two models are reported in table 3.7 The first includes all
individual-level predictors, along with the CPI scores for each nation.
The second adds CPI
political interest and CPI
education interaction terms to test whether the similarity between CPI scores and mass
perceptions of corruption peaks for certain respondents. In pooled

100

LATIN AMERICAN POLITICS AND SOCIETY

47: 3

cross-national analyses such as this, data are inherently clustered by

nation in that they are gathered in separate surveys conducted in each
of several countries. Failure to account for this clustering can result in
unduly small standard errors and produce too liberal of significance
tests for coefficient estimates. To avoid this problem, these and subsequent models are estimated using robust standard errors that account for
the national-level clustering in these data.
Results in table 3s initial model reveal that several individual-level
variables correspond with the perception of corruption. Specifically,
women, older respondents, and respondents who lack interpersonal
trust and political interest all tend to perceive higher rather than lower
levels of political corruption. Even after controlling for the impact of
these variables, though, an extremely strong effect is found for the CPI
variable. The coefficient on the Transparency International indicator is
nearly eight times the size of its standard error, reaching the highest
level of statistical significance. Citizens views of corruption in Latin
America apparently are neither unfocused nor unfounded. To the contrary, there is a high degree of correspondence between expert judgments and the views of the person on the street.
To be clear, it is not our claim that the typical citizen learns of the
CPI and uses those data to inform perceptions about levels of corruption. Instead, we use CPI scores to approximate variance in actual levels
of corruption in the six nations under consideration. From this perspective, the large and statistically significant coefficient on the CPI variable
suggests that the reality of political conditions (as captured by elite
perceptions) in these nations resonates with the mass public. Students
of political behavior often are concerned about the consequences that
emerge when citizens are less than fully informed about politics.
Although it is admittedly a stretch to argue that the CPI data measure
actual levels of political corruption, these data do at least indicate corruption as perceived by attentive, well-informed observers. From this
perspective, the significant effect for the CPI variable reveals that mass
opinion about corruption is neither haphazard nor baseless.
Although the initial model in table 3 demonstrates that mass judgments regarding levels of corruption vary in response to levels of corruption as indicated by the CPI, a limitation of this specification is that
it fails to recognize possible variance among mass judgments. We have
speculated that the connection between CPI data and mass perceptions
should be strongest for those citizens who have the resources and
capacity to monitor the political scene most closely. In other words,
some citizens may be better positioned than others to perceive corruption accurately. This hypothesis is tested by determining whether the
impact of corruption as indicated by CPI data is moderated by two factors critical to political engagement: education and interest in politics.

CANACHE AND ALLISON: CORRUPTION

101

Figure 2. Moderating Impact of Political Interest on

Perception of Corruption

Statistically significant positive coefficients on the interaction terms CPI

education and CPI
political interest would establish the existence
of the hypothesized conditional effects.
In column 3 of table 3, we see a strong and statistically significant
interaction for interest, although not for education. Consistent with
expectations, the interest effect suggests that citizens who are most
attuned to politics are most capable of gauging the extent to which corruption pervades the political system. Figure 2 graphs the substantive
impact of the political interest and CPI interaction. Specifically, the
figure displays the estimated likelihood that a respondent perceives all
or most public officials, as opposed to only a few or almost none, to be
corrupt; variables other than the CPI score and political interest are held
constant. Where corruption is high, at least as indicated by high values
on the CPI, all respondents perceive it as such, irrespective of variance
in their levels of interest in politics. This is a plausible result. In nations
that are absolutely rife with political corruption, the signs will be visible
even to citizens who follow politics only sporadically.
It is in Latin American nations where corruption is relatively light
nations such as Chile and Uruguaythat variance in political interest

102

LATIN AMERICAN POLITICS AND SOCIETY

47: 3

appears to matter the most. Respondents who are uninterested in politics have a 50-50 chance of saying that all or most leaders are corrupt,
versus a mark of less than 20 percent for respondents who are highly
interested in politics. What this suggests is that political interest functions
to permit incumbent officials from standing falsely accused. To take one
example, Chileans with high levels of political interest are much more
likely than their politically disinterested counterparts to recognize that
Chile is not suffering from a widespread epidemic of political corruption.
Collectively, these results demonstrate that Latin Americans do
indeed recognize corruption in their nations political systems. If anything, corruption is overstated by politically disinterested respondents in
nations where corruption is not so severe. The important point here is
that corrupt leaders arent fooling anyone. Citizens of these nations see
the problem. This is the first step toward accountability. The second
step is for citizens to link their perceptions of corruption to appraisals
regarding incumbent officials and the political system. Whether this
second step occurs is the next question.

PERCEIVED CORRUPTION AS A DETERMINANT

OF SUPPORT FOR INCUMBENTS AND INSTITUTIONS
This study adopts a hierarchical approach to the study of political support in which each level of support represents a higher level of generalization than its immediate predecessor (Canache 2002). Political support is treated as a multidimensional construct incorporating
components at three different levels of generalization. At the least general level, citizens assess the performance of incumbent officials,
whereas at the most general level, evaluations center on the type of
regime operating in the nationdemocracy, in the case of the nations
currently under consideration. At the intermediate level, citizens form
opinions regarding the political institutions of the regime, such as the
courts, the military, the police, and the bureaucracy.
Perceptions of corruption may influence any or all of the three levels
of political support. The wisdom of linking views on corruption to other
political evaluations arguably varies, however, depending on the judgment at hand. If incumbent officials are corrupt, it makes perfect sense
for citizens to evaluate them negatively. Likewise, if a nations government institutions are riddled with corruption or if political procedures
facilitate corrupt practices, citizens are right to take a critical view. Matters differ, however, when we come to the question of regime support.
High levels of political corruption are not endemic to democracy. After
all, corruption is much less severe in Chile, Costa Rica, and Uruguay than
in Colombia, Mexico, and Venezuela, and corruption is virtually nonexistent in nations such as Finland and New Zealand. If the perception of

CANACHE AND ALLISON: CORRUPTION

103

high levels of corruption leads citizens to form negative views of democracy itself, then we would have something of a baby with the bath
water phenomenon. Corruption would impose a severe cost if it led citizens to view democratic governance with indifference or hostility.
The assessment of the political consequences of corruption begins
by exploring whether the perceptions discussed above influence levels
of support for incumbent officials and a nations political system. Some
evidence already suggests that citizens do link corruption to opinion
about political actors. Specifically, Morris (1991) identified a strong tie
between Mexicans perceptions of corruption and their levels of trust in
government. Seligson (2002), using a measure of actual experience with
corruption, reports strong relationships between the corruption variable
and system support in Bolivia, El Salvador, Nicaragua, and Paraguay.
In the analyses for this study, citizen support for the incumbent government is measured with data from a four-category ordinal item that
asked respondents to indicate how satisfied they were with the way the
national government is addressing the countrys affairs. Citizen support
for the political system is operationalized using a 15-point scale constructed with data from a series of questions on which respondents were
asked to rate their confidence in several political institutions: the legal
system, the government, the police, the parliament, and the civil service.
Support for the incumbent government is modeled using ordered logistic regression, whereas an OLS model is estimated for system support.
The primary independent variable is the four-point indicator of perceived political corruption; but the models also include several additional predictors. Among these are various sociodemographic indicators
(sex, age, education, and social class), sociocultural orientations (life satisfaction and financial satisfaction), and political orientations (political
interest, views on paternalism and government liability). The items used
to construct these variables are reported in the appendix, along with
each variables descriptive statistics.
The regression results depicted in table 4 reveal that several of the
control variables exert significant effects on political support. Of primary
interest, of course, is the impact of perceived corruption. Statistically
significant coefficients are obtained for the corruption variable in both
models. These results establish that the second half of the accountability equation is in place. When citizens in Latin America perceive rampant political corruption, those views translate directly into a downgrading of opinion for both incumbent officials and political institutions.
When elected officials engage in corrupt practices, they do so at their
own political peril.
Again, it is all to the good if people correctly perceive corruption
and then transfer those perceptions into negative evaluations of incumbent officials and even features of a nations political system, but it

104

LATIN AMERICAN POLITICS AND SOCIETY

47: 3

Table 4. Effects of Perceived Corruption on Political Support

Support for
Incumbent
Support for
Government
Political
(ordered logistic
System (OLS
regression estimates) regression
estimates)
___________________
__________________
Robust
Robust
Standard
Standard
Coefficient Errora Coefficient Errora
Sex
Age
Education
Social class
Satisfaction with life
Satisfaction with financial
situation
Political interest
Views on paternalism
Evaluation of poverty
situation
Evaluation of government
action toward poverty
Perceived corruption
Constant
Number of observations
Model 2
Model R2

0.058
0.009*
0.046#
0.100
0.008#

0.087
0.004
0.026
0.068
0.004

0.057
0.010
0.116*
0.114
0.007

0.063
0.010
0.033
0.080
0.006

0.083***
0.167*
0.051#

0.023
0.075
0.028

0.150*
0.543**
0.046*

0.041
0.116
0.017

0.560***

0.127

0.335

0.170

0.819***
0.344**

0.164
0.122

1.023**
0.816**
10.996***

0.217
0.184
0.694

5,203
1,203.33

5,127
0.146

***p < .001 **p <.01 *p <.05 #p < .10

a
Adjusted for clustering on country.
Dependent variables: support for the incumbent government, a four-category ordinal measurement; and support for the political system, an interval measurement.
See notes to table 3.
Sources: World Values Survey 199597; Transparency International.

would be troubling if views on corruption caused citizens to turn their

backs on democracy. All the countries in our sample can be characterized to a certain extent as fragile democracies. In such nations,
although formal democratic procedures may have been installed, democratic roots are not well entrenched, and therefore significant levels of
uncertainty about the sustainability of democracy exist (Canache 2002,
6). Although corruption in advanced democracies is unlikely severely to
erode public support for democracy as the most desirable form of gov-

CANACHE AND ALLISON: CORRUPTION

105

Table 5. Effect of Perceived Corruption on Support for Democracy

(OLS regression estimates)

Sex
Age
Education
Social class
Satisfaction with life
Satisfaction with financial situation
Political interest
Views on paternalism
Evaluation of poverty situation in
the country
Evaluation of government action
toward poverty
Perceived corruption
Constant
Number of observations
Model R2

Coefficient

Robust
Standard Errora

0.035
0.007#
0.048*
0.024
0.001
0.011
0.081
0.003

0.035
0.003
0.017
0.051
0.008
0.014
0.045
0.008

0.067

0.067

0.201*
0.029

0.054
0.035

4.067***

0.129

4,932
0.029

***p < .001 **p <.01 *p <.05 #p < .10

aAdjusted for clustering on country.
Dependent variable: Support for democracy, an interval measurement. See notes to
table 3.
Sources: World Values Survey 199597; Transparency International.

ernment, there is less reason to expect the same unwavering commitment to democracy in these Latin American countries. We know that
more and more Latin Americans are ambivalent or genuinely hostile to
democracy as a form of government. For instance, data from the 2002
Latinbarmetro show that support for democracy in the region as a
whole was only 56 percent (Latinobarmetro 2002). We also know that
opposition to democracy as a form of government is associated with
both support for political violence and support for radical leadership
alternatives. Thus, the very real possibility exists that political corruption
in Latin America poses a tangible threat to democratic stability.
Data from two items are combined to form an indicator of support
for democracy, yielding a scale with values ranging from 2 (low support) to 8. This scale is the dependent variable in table 5. The OLS
regression results reveal no evidence that perceptions of corruption
undermine support for democracy as a form of government. This impor-

106

LATIN AMERICAN POLITICS AND SOCIETY

47: 3

tant null result speaks to an element of rationality in citizens political

appraisals, especially when viewed in the context of the significant
effects reported in table 4 for this same variable. In Latin America, citizens make a critical evaluative distinction. When they perceive political
corruption to be widespread, they lash out at both incumbent officials
and national political institutions, but, at least thus far, that is where they
draw the line. Perceptions of political corruption apparently do not
undermine support for democracy as a form of government.

CONCLUSIONS
Mass opinion potentially offers one check on corruption in that a
nations citizens could demand higher standards of integrity from public
officials. For this sort of accountability to occur, however, citizens must
recognize corruption when it exists, and evidence of corruption must
influence citizens judgments regarding the culpable parties. The central
objective in this article has been to determine whether accountability of
this form does indeed take place in Latin America. A secondary objective involved ascertaining whether perceptions of corruption cause citizens to look critically at all aspects of democratic politics, thereby contributing to an erosion in support for democracy in the region.
Drawing on Transparency Internationals corruption index (CPI)
and World Values Survey (WVS) data from several Latin American
nations, we have tested a series of empirical relationships regarding perceptions of corruption and political attitudes. Collectively, these findings
suggest that the necessary ingredients for accountability are present in
Latin America. Citizens perceive corruption, and they connect those perceptions to their judgments of incumbent leaders and political institutions. Whether mass opinion alone can offer a sufficient check on corruption is, of course, highly questionable. At a minimum, however, it
surely is better to have accountability of the sort identified here than to
see corruption flourish unhampered by public criticism.
The possible dark side of mass opinion regarding corruption is that
pervasive and often burgeoning misconduct may poison public sentiment toward democratic politics in general. On this score, these analyses speak somewhat to the vitality of democracy in the region, at least
in the short term. Although many Latin American systems are rife with
corruption, this situation only seems to affect support for specific administrations and institutions. This analysis identified no evidence that perceptions of corruption have soured mass opinion on democracy as a
form of government. Unfortunately, it would be perilous to assume that
this state of affairs can continue indefinitely. Like Seligson 2002, this
study found that corruption undermines the legitimacy of governments
in Latin America; and like Seligson, we suspect that democracy in the

CANACHE AND ALLISON: CORRUPTION

107

region, as a consequence, may be racing against the clock. Citizens in

the region apparently do not currently view corruption as an inherent
feature of democratic governance, but there is no guaranty that this perceptual link will not emerge in future years if corruption continues
unabated. Given that democracy already is on unsure footing in many
Latin American nations, concern is warranted that the persistence of corruption eventually may add to the allure of nondemocratic regimes.

APPENDIX: WORDING
STATISTICS

AND

DESCRIPTIVE

Table 6. Wording and Descriptive Statistics of Survey Questionnaire

Variable

Scale (lowesthighest)

Statistics

Sex

1 = male; 2 = female

Mode = 2

Age

Years (1891)

Mean = 39.01
SD = 15.03

Education
1 = no formal
What is the highest educaeducation
tional level you have attained? 9 = university degree

Mean = 5.04
SD = 2.26

Social Class
1 = lower class
People sometimes describe
5 = upper class
themselves as belonging to the
working class, the middle class,
or the upper or lower class.
To which would you describe
yourself as belonging?

Mean = 2.55
SD = 0.88

Satisfaction with life

All things considered, how
satisfied are you with your
life as a whole these days?

1 = completely
dissatisfied
10 = completely
satisfied

Mean = 7.24
SD = 2.48

Satisfaction with financial

situation
How satisfied are you with
the financial situation in your
household?

1 = completely
dissatisfied
10 = completely
satisfied

Mean = 5.89
SD = 2.77

Political Interest
How interested would you
say you are in politics?

1 = not at all
interested
4 = very interested

Mean = 1.98
SD = 0.98

108

LATIN AMERICAN POLITICS AND SOCIETY

47: 3

Table 6. (continued)
Variable

Scale (lowesthighest)

Statistics

Interpersonal Trust
Generally speaking, would
you say that most people can
be trusted or that you cant be
too careful in dealing with
people?

0 = Cant be too
careful
1 = Most people
can be trusted

Mode = 0

Views on Paternalism

1 = People should
take more responsibility to provide for
themselves
10 = The government
should take more
responsibility to
ensure that everyone
is provided for

Mean = 6.07
SD = 3.13

Evaluation of Poverty Situation 1 = a smaller share

Would you say that today a
3 = a larger share
larger share, about the same,
or a smaller share of the
people in this country are
living in poverty than were
ten years ago?

Mean = 2.62
SD = 0.65

Evaluation of Government
Action Toward Poverty
Do you think that what the
government is doing for
people in poverty in this
country is the right amount,
too much, or too little?

1 = too little
3 = too much

Mean = 1.29
SD = 0.54

Transparency International
Score (CPI index)

0 = low-level
corruption
10 = high-level of
corruption

Mean = 6.54
SD = 1.21

Perceived Corruption
How widespread do you
think bribe taking and corruption are in this country?

1 = Almost no public
officials are engaged
in it
4 = Almost all public
officials are engaged
in it

Mean = 3.02
SD = 0.88

CANACHE AND ALLISON: CORRUPTION

109

Table 6. (continued)
Variable

Scale (lowesthighest)

Support for the Incumbent

1 = very dissatisfied
Government
4 = very satisfied
How satisfied are you with
the way the people in national
office are handling the
countrys affairs?

Statistics
Mean = 2.12
SD = 0.90

Support for the Political System

Scale combining items gauging
the level of confidence in the
following institutions: legal
system, government, police,
parliament, civil service

5 = low level of
confidence
20 = high level of
confidence

Mean = 10.99
SD = 3.67

Support for Democracy

Scale combining the following
items:

2 = low support
for democracy
8 = high support
for democracy

Mean = 3.52
SD = 1.25

Having a democratic system

is a very good, fairly good,
fairly bad, or very bad way of
governing this country? (1 =
very bad, 4 = very good)
Democracy may have problems but its better than any
other form of government.
(1 = disagree 4 = agree)
Sources: World Values Survey 199597; Transparency International.

NOTES
1. For an excellent review on the economic and political literature on corruption, see Seligson 2002.
2. These include Alan Garca (198590) in Peru; Fernando Collor de Mello
(199092) in Brazil; Carlos Salinas de Gortari (198894) in Mexico; Abdal
Bucaram (199697), Fabin Alarcn (199798), and Jamil Mahuad (19982000)
in Ecuador; Ernesto Samper (199498) in Colombia; Carlos Menem (198999) in
Argentina; and Alberto Fujimori (19902000) in Peru.
3. The scale of the original CPI index ranges from (0) high corruption to
(10) low corruption. Here the scale is inverted for the sake of comparability and
interpretation of the variable used in subsequent analyses. For further concep-

110

LATIN AMERICAN POLITICS AND SOCIETY

47: 3

tual and methodological issues regarding the CPI index, see Graf Lambsforff
2003.
4. For additional discussion of the strengths and limitations of the CPI, see
Seligson 2002. We share many of Seligsons concerns regarding use of CPI data
in individual-level analyses. Like Seligson, however, we also think these data
offer a barometer against which to gauge the alternate measures we devise.
5. Another cross-national survey, the Latinbarmetro 1997, asked respondents to assess corruption in their nations. But the items did not directly measure respondents perceptions regarding levels of corruption, and thus they are
not directly useful for the research questions examined here. In our view, the
World Values Survey items provide a more straightforward way to capture how
much corruption the respondents think exists in each country. In 2002, Transparency International launched the Global Corruption Barometer, a new initiative aiming to monitor attitudes and perceptions about corruption on a global
scale. This new effort promises to offer an important reference for future
research on mass perceptions of corruption.
6. Seligson (2002, 421) examines the relationship between Transparency
International data and survey data on corruption experienced by survey respondents in Bolivia, El Salvador, Nicaragua, and Paraguay and, as this study does,
finds a very high level of correspondence between the CPI and survey-based
indicators.
7. These models include data from only six nations because the items used
for two of the independent variables were not asked in Colombia.

REFERENCES
Banfield, Edward. 1958. The Moral Basis of a Backward Society. Chicago: Free
Press.
Becquart-Leclerq. 1989. Paradox of Political Corruption: A French View. In Heidenheimer et al. 1989.
Camp, Roderic Ai, Kenneth M. Coleman, and Charles L. Davis. 2000. Public
Opinion About Corruption: An Exploratory Study in Chile, Costa Rica, and
Mexico. Paper presented at the annual meeting of the World Association for
Public Opinion Research, Portland, Oregon. May 1718.
Canache, Damarys. 2002. Venezuela: Public Opinion and Protest in a Fragile
Democracy. Coral Gables: North-South Center Press.
Della Porta, Donatella. 2000. Social Capital, Beliefs in Government, and Political
Corruption. In Dissatisfied Democracies: Whats Troubling the Trilateral
Countries? ed. Susan J. Pharr and Robert D. Putnam. Princeton: Princeton
University Press. 20228.
Diamond, Larry, Andreas Schedler, and Marc F. Plattner. 1999. Introduction. In
Schedler et al. 1999. 110.
Etzioni-Halevy, Eva. 1985. Bureaucracy and Democracy. London: Routledge
and Kegan Paul.
Graf Lambsdorff, Johann. 2003. Background Paper to the 2003 Corruption Perception Index (Transparency International 2003). <www.transparency.org>
Heidenheimer, Arnold J., Michael Johnston, and Victor T. LeVine, eds. 1989.
Political Corruption: A Handbook. New Brunswick: Transaction.

CANACHE AND ALLISON: CORRUPTION

111

Huntington, Samuel P. 1968. Political Order in Changing Societies. New Haven:

Yale University Press.
Johnston. Michael. 1979. Patrons and Clients, Jobs and Machines. American
Political Science Review 73: 38598.
Latinobarmetro. 2002. Press Report: Latinobarmetro 2002 Public Opinion Poll.
<www.latinobarometro.org>
Morris, Stephen D. 1991. Corruption and Politics in Contemporary Mexico.
Tuscaloosa: University of Alabama Press.
ODonnell, Guillermo. 1999. Horizontal Accountability in New Democracies. In
Schedler et al. 1999. 2951.
Pharr, Susan J. 2000. Officials Misconduct and Public Distrust: Japan and the Trilateral Democracies. In Dissatisfied Democracies: Whats Troubling the Trilateral Countries? ed. Susan J. Pharr and Robert D. Putnam. Princeton:
Princeton University Press.
Schedler, Andreas, Larry Diamond, and Marc F. Plattner, eds. 1999. The SelfRestraining State: Power and Accountability in New Democracies. Boulder:
Lynne Rienner.
Seligson, Mitchell A. 2002. The Impact of Corruption on Regime Legitimacy: A
Comparative Study of Four Latin American Countries. Journal of Politics 64:
40833.
Shin, Doh C. 1999. Mass Politics and Culture in Democratizing Korea. Cambridge: Cambridge University Press.
Stapenhurst, Rick. 2002. The Media Role in Curbing Corruption. Washington,
DC: World Bank Institute.
Transparency International. 1997. Corruption Perception Index 1997.
<www.transparency.org>
. 2003. Corruption Perception Index 2003. <www.transparency.org>
Tulchin, Joseph S., and Ralph H. Espach, eds. 2000. Combating Corruption in
Latin America. Washington, DC: Woodrow Wilson Center Press.
Waterbury, John. 1976. Corruption, Political Stability, and Development: Comparative Evidence from Egypt and Morocco. Government and Opposition
11: 42645.
Weyland, Kurt. 1998. The Politics of Corruption in Latin America. Journal of
Democracy 9: 10821.
World Values Survey (WVS). 199597. Ann Arbor: ICPSR, University of Michigan.
<nds.umdl.umich.edu>